Author(s)

Bailin Pan, Huan Le, and Yumin Wang

Corresponding Author

Bailin Pan

Abstract

Taking Hangzhou Cigarette Factory of Zhejiang Zhongtobacco Industry Co., Ltd. as an example, Aiming at the back flow fault of finished tobacco scanning and sorting in the cigarette factory's packaging logistics production line, through the construction of the auxiliary system for the back flow fault tracing of finished tobacco scanning and sorting, the algorithm analysis of the relevant indicators of the production environment is carried out, the alarm convergence ability is improved, the root cause of the fault is quickly located, the steady-state model of the production business is established, and the fault tracing function is realized. Improve the ability of production data visualization, production process transparency and production decision-making intelligence. The back flow fault of finished tobacco code scanning and sorting is caused by the jitter of environmental factors caused by PLC transmission mechanism. In this study, various monitoring indicators related to production environment are collected through various technical tools, such as ZABBIX, APM, network packet capturing, application buried point monitoring, etc., which are usually stored as time series data (including collection time and indicator value), and a large number of indicators will be collected The historical data and real-time data of are imported to the big data platform to clean, store, analyze and mine the environmental data related to the production process of finished tobacco scanning and sorting. At the same time, an end-to-end model that can reflect the relationship between topology strength and weakness is constructed through the application performance database of traceable volume package logistics combined with the business architecture for visual display. It provides ideas for the intelligent construction of operation and maintenance of production system in cigarette industry.

Keywords

Time Series Data, Association Algorithm, Anomaly Location, Semi Supervised Machine Learning, Fault Backtracking